solaris agent

[zabbix_gunner]

am running zabbix agent on solaris 11 SPARC. I see a strange issue. Every so often the server shows the dreaded "timeout" error. Restarting the agent resolves this. It has come down to restarting the agent once a minute to collect all data.
attaching agent logs when the server shows "timeout". Nothing seems to be out of the ordinary there.

any pointers on this ?. Has anybody run into this ?

regards

AK

[tchjts1]

There is a Timeout= value on the Zabbix App server in zabbix_server.conf as well as on your monitored host in zabbix_agentd.conf

If those are at the default settings, you can try bumping them up to see if that resolves the issue. On Zabbix server, I like to put that up to 15 from the default of 3. If you do make changes, just put in a new line below the default line without a comment "#" preceding them or Zabbix will still use the default setting.

If you make changes to either of those files, you will need to restart the corresponding procs.

Also, have you checked your graphs for Zabbix internal items to see how the performance is for them? Refer to this post: https://www.zabbix.com/forum/showthread.php?t=47781

[zabbix_gunner]

sorry , forgot to mention. I had already increased the T/O value at both the server and agent level to 30 seconds each. Also zabbix internal graphs are pretty much as they were before this agent was added.

regards

AK

[tchjts1]

Solaris can be a pain sometimes. Is your version of Zabbix server and agent a match?

I am not saying that the agent is the issue, but if it is the only monitored host you are having issues with, you may consider playing with that to see if it resolves the problem.

The times that I have run into Solaris issues with a pre-compiled agent from the Zabbix download site, I would try downgrading the agent version first to see if that helped. If not, then I would simply put the Zabbix source code onto that particular Solaris server and custom compile the agent. Takes like 5 minutes to crank out a new binary that way.

If that does not fix the issue, at least you have ruled out the agent as the culprit and I would start looking at possible network issues with that box.

I don't know if you are using passive or active agent items for your monitored items, but if they are mostly passive, I would try active and see if that helps as well.

[kloczek]

Solaris can be a pain sometimes. Is your version of Zabbix server and agent a match?

I am not saying that the agent is the issue, but if it is the only monitored host you are having issues with, you may consider playing with that to see if it resolves the problem.

On every OS exact version of agent is working with even updated versions of servers/proxies.
List of supported versions of the agents is in documentation of exact version of server/proxy.
It has nothing to do with Solaris.

In attached agent log are only entries confirming that agent has nothing to send because there is no any monitored items on this host in zabbix setup.

[tchjts1]

On every OS exact version of agent is working with even updated versions of servers/proxies.
List of supported versions of the agents is in documentation of exact version of server/proxy.

That is your opinion. And you are welcome to have it.

[bbrendon]

I would try an older agent. I have found many times in the past there is a sweet spot for agents and OS distros.

This is likely possible with Solaris. I know bugs get introduced in newer agents that go unresolved for years and only the mainstream agents get much attention.

I found a v1.8 agent running on a system a few months ago and our server is v2.4 so pretty much any agent should work.

[zabbix_gunner]

On every OS exact version of agent is working with even updated versions of servers/proxies.
List of supported versions of the agents is in documentation of exact version of server/proxy.
It has nothing to do with Solaris.

In attached agent log are only entries confirming that agent has nothing to send because there is no any monitored items on this host in zabbix setup.

what do you mean by "no any monitored items on this host" ? the solaris templates have been attached to the host.

My agent version is 2.2.0 and server version is 2.2.4. So that should not be too much of a problem from that pov.

For now , i have set up a cronjob to restart the agent every minute ( yes that is how bad it is ). I guess I will try another version of the agent and see if that solves the issue.

regards

AK

[zabbix_gunner]

used agent version 2.2.5 and that seems to have done the trick... more than 3 hours w/o restart and data is being collected seamlessly...

regards

AK

[kloczek]

what do you mean by "no any monitored items on this host" ?

In sample log there is no any entries sending even one sample of collected data.
Possible explanations are two:
- no monitored items
- sample log is to short to indicate that some data have been collected.

[zabbix_gunner]

In sample log there is no any entries sending even one sample of collected data.
Possible explanations are two:
- no monitored items
- sample log is to short to indicate that some data have been collected.

there are monitored items on the host.

log stays the same when data is being sent. debug 4

[bunina1]

I would try an older agent.

[zabbix_gunner]

already done and it is works... see an earlier post in thread

[kloczek]

already done and it is works... see an earlier post in thread

It must be coincidence that it works for you after downgrade because agent 2.4.5 is working perfectly fine.

Again: in attached by you sample log there is no single logged error of any entry showing that something is wrong or you did not present what is wrong or what is the problem.

[zabbix_gunner]

1./was not exactly a downgrade.. instead of 2.2.0 i used 2.2.5 and it worked... dont believe in coincidences... could be an issues with ver 2.2.0

2./ as i already said , the log looks the same when the data was not being sent and now that everything is fine...